INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Gra

-0.80

-+

-0.76

Answer

-0.75

Tow

-0.73

Upload

-0.68

Sound

-0.67

 nutshell

-0.67

BIL

-0.65

 ********************************

-0.64

************

-0.64

POSITIVE LOGITS

ypes

0.81

caster

0.74

zynski

0.74

oen

0.70

nance

0.68

metry

0.67

iatrics

0.67

acters

0.67

mite

0.67

casters

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.